Launching Pretty Prompt - The Real Story.
From 0 to 10k+ improved prompts from Product Hunt!
How it all started
A couple of weeks ago, while building our first product, Dolphin AI, we kept coming up with a blocker. We kept fighting with AI to get the AI to do what we needed to do.
Prompt Engineering is hard!
We realized that building a great product is half engineering, half prompting. At least in 2025…
But this is the first time we’ve had to learn how to talk to an LLM.
Writing good prompts is weirdly hard, and refining them? Even worse. It's a constant battle of iteration, just like building a new product.
So we did what any founder would do - we went ahead and built a tool to solve our own problem. We called it Pretty Prompt.
Over a weekend, we built a (buggy) MVP, just for ourselves. What had previously taken constant back and forth was now just a button away. I loved it. Gave us the ability to get 10x out of AI.
Fast forward a couple of weeks, Pretty Prompt has been used in more than 10,000 prompts, installed over 4,000 times, and creators are making TikToks of it!
I don’t even have TikTok myself… I guess I’ll have to open an account 😅.
This was enough validation to build something properly. And we’re sharing our journey while doing it. Want the short version? Click here.
This is how we went from zero to paying users in under a week.
Hope you enjoy it. Give Pretty Prompt a go! It’s pretty cool ✨.
Validation 0: Solving our own problem first
I’m convinced that one of the best ways to build a startup is to build something for yourself first. Not as a startup. But as a project.
Something fun, something you really want. Something that helps you fix that one thing in your day-to-day. That one thing that blocks you from moving forward, or that’s simply too annoying to do.
For us, it was Prompt Engineering. It was something that we had to fix or tweak every single day.
So we took the challenge to build a new tool to fix it.
The v0 of Pretty Prompt was done completely through Vibe Coding (thank you, Base44.)
It wasn’t great. It wasn’t what we actually wanted. But it proved one thing: It felt magical.
That was our validation. Our version zero.
Before building a product for the world, we had to answer:
Would we use this? Would we want to pay for it? The answer was YES and YES.
So over a weekend, Charlie coded an MVP, to share with the world.
There was no crazy scope. No big strategy or design. Just a simple Notion page that said the following:
48 hours of work, and our first learning was that building a Chrome Extension is quite different from building a web app. More on this later…
The outcome?
A functional MVP. A Product Hunt launch. 2nd Product of the day. And the conviction that there was something special in Pretty ✨.
Validation 1: Product Hunt Launch. From Scrappy Idea to "Wait, What?! People want to pay for this?"
There was no fancy pitch deck, no long-term plans – just a scrappy MVP we wished existed.
Pretty Prompt was like Grammarly, but for prompting.
Neal Travis from Growth Support joined the team to help with GTM, Content, and Ops.
We scheduled the launch on Product Hunt to go live on May 31, 2025. (Btw, this is Charlie’s birthday 🎉…)
We didn’t put much effort into the launch. We even forgot it was going live that weekend. But Neal gave us all a Startup masterclass with his message:
“It could be a good chance to pressure test it and see”. - Neal Travis.
A launch is not important. It is what happens after it. Does anyone even remember when Shopify launched? Or when Airbnb did? Nope. Here’s a nice tweet by Brian Chesky on launching multiple times.
-
Let’s just flow with it, and see what happens…
So we let it be. What happened next?
An explosion. Seriously, it went crazy!
Within just a few days, the app was getting thousands of installs and improving thousands of prompts. We were getting emails saying:
“Hello I was trying to subscribe and potentially pay for the service but I have had no luck getting to a payment page.”
There was no paywall! 🫣
-
This wasn't just our pain. A lot of people were feeling it, and they were hungry for a simple solution. Happy to pay for it. And happy to give feedback to make it even better.
The market always wins. And when you feel that pull, you need to give it everything you’ve got.
Getting people to pay would be the next milestone…
Validation 2: Paywall, feedback, and fast iteration
So after Product Hunt blew up, and we didn’t have a paywall, we took a few steps to move forward with Pretty:
Answer to every single comment on Product Hunt.
Search for every newsletter or website where people had mentioned Pretty Prompt, to engage with them, and say thank you.
Add a paywall. (Kind of backwards, but hey, that’s life!)
We added a simple Stripe checkout, with one plan, simple pricing, and suddenly…
People started to pay. 🥹
And started to leave powerful reviews on Chrome.
It wasn’t some huge marketing move. It was word of mouth.
Someone finds a tool they love → Shares it with a friend → And before you know it, you’re waking up to Stripe notifications. And of course, requests, feedback, and bugs.
All part of the game.
YC’s motto still holds: “Build something people want.”
Something we did pretty well over the past year while building Dolphin was the speed of execution. Speed compounds over time. We’re pushing ourselves to keep this with Pretty. But Chrome Extensions are slightly different from building a web app.
Shipping updates to a Chrome Extension isn’t as instant as with your own application.
You can’t just push to prod and see it live in 3 minutes.
Chrome needs to review and approve every update. And at first, this feels annoying.
We ship daily! Why do we need to wait for approvals!?
But then I realized — people aren’t sitting around refreshing your extension every hour. They have jobs. Families. Netflix.
When was the last time you got a proper update from LinkedIn? Exactly.
So, even the right improvements every 3 days are faster than most.
That’s what we’re aiming for.
(3-5 days is usually the review cycle on Chrome.)
Organic growth also compounds.
Every morning, I would Google “Pretty Prompt” and find it somewhere new.
TikToks, newsletters, LinkedIn posts, and AI communities. (Pro tip - Set it up as a task on ChatGPT.)
People have been sharing it because it just works.
I’m convinced that the best thing that has ever happened to a startup is having early adopters. They are the superheroes of the tech industry 🦸♂️.
There are probably many reasons why this has worked much better than Dolphin AI, so far.
One of those could be because Pretty Prompt is a Chrome Extension. Not everyone likes them, but Chrome Extensions are:
Specific.
Contained and, in effect, simpler.
Lightweight.
Some examples of apps I love: Jam for bug reports. Grammarly for typos. 1Password for logins.
Pretty Prompt for prompting.
That was the magic. Simple, meets you where you already are, no fluff.
Another thing we did, once the paywall was up, was to embrace building in public and sharing it with our network.
Here’s my launch LinkedIn post. (21k impressions, 222 reactions, 125 comments.)
What’s Next for Pretty Prompt
We’re listening like crazy, shipping daily, and fixing every bug possible, to make the experience as good as if you were on Instagram or Notion.
We’re still in the early days. But here’s what’s coming:
Making the app more robust and reliable.
Snooze the app for focus mode.
Adding multiple languages. So that when you prompt it Spanish, French, Hebrew (or other languages), the output is in the same language by default.
Adding more platforms: ChatGPT, Gemini, Claude, Desktop, etc.
Team plans so you can manage prompts together.
And more. There’s a lot to do. And as I used to say in Spanish, “Esto es solo el comienzo…” (This is just the beginning). I’m from Argentina 🇦🇷.
—
We built this for ourselves because we were sick of fighting with prompts. Now, it's yours too.
We're constantly improving and always listening.